ACTA LINGUISTICA LITHUANICA

LIII (2005), 33–37

Remarks on Winter’s Rule

Witold Mańczak

Cracovie

The author presents a few observations concerning Winter’s Rule, among which the most important are the following: (1) the author knows of no language in which a phonetic change similar to that ascribed to Balto-Slavic by Winter would have occurred; and (2) certain exceptions to Winter’s Rule could be explained by what the author calls irregular phonetic development due to frequency.

1. There is a fundamental difference between the observation that Napoleon died in 1821 on Saint Helena and the observation that man (or, more generally, every living being) is mortal. The first observation, limited in time and space, applies to a historical event, whereas the latter, valid always and everywhere, is a biological law. Mutatis mutandis, there are similar differences in linguistics. Roman grammarians knew only one rule of historical phonetics: vocalis ante vocalem corripitur (for example, habēre, habēm̄us, habētis, habēbat but habeō, audīre, audīmus, audītis, audītum but audiō, audiunt, audiēbat). The rule in question concerns a single language (Latin), whereas Zipf’s law applies to all languages and to all periods of their history. Since there are at least 3,000 languages in the world, the difference between the rule vocalis ante vocalem corripitur and Zipf’s law is enormous; it is comparable to the difference between 1 and 3,000. Given that “Winter’s law” concerns only one language (Balto-Slavic) and therefore resembles the rule vocalis ante vocalem corripitur, and not Zipf’s law, it is more accurate to call it a “rule.” The fact that the vast majority of linguists confuse laws and rules is explained by their complete lack of interest in the quantitative aspect of linguistic facts and their failure to see any difference between 1 and 3,000.

2. Winter (1978: 444–445) wrote the following:

“Is a development of the type VD > V:D a natural one? Or is what we seem to find in Baltic and Slavic languages an unusual phenomenon?

Rather than attempt to give a long list of parallels from other languages, I will limit myself to a brief discussion of a well-known development within Polish.

In a very large number of Polish words, we find an alternation of the type dwór, dworu... The change, which from present conditions would have to be called a vowel raising, is interpreted as vowel lengthening... The conditions for the introduction of length seem to have been as follows: a short o, whether derived from older o or older e, was

34 |

W I T O L D M A Ń C Z A K

lengthened in a position before voiced nonnasal consonant, provided this consonant was followed (in terms of present-day Polish) by word boundary or morphonic boundary plus stop. It is interesting to note both similarities and differences between the conditions for lengthening in medieval Polish and in Balto-Slavic: For Polish, the decisive fact seems to be that a consonant have the feature [+ voiced], for Balto-Slavic, that this feature be a distinctive one – thus lengthening would occur before r in medieval Polish, but not in early Balto-Slavic. In Polish then, the phenomenon has a wider range as far as the immediately following phonemes are concerned; on the other hand, there is no requirement in Balto-Slavic that the phoneme triggering the lengthening process occur in a specific environment. Lengthening in medieval Polish and in early Balto-Slavic are, therefore, similar, but by no means identical, processes; there was in medieval Polish no revival, as it were, of an ancient rule but similar conditions merely led to similar results.”

There is an essential difference between vowel lengthening in Old Polish and in Balto-Slavic. In Old Polish, this was a change similar to what is currently happening in French ruse compared with russe or in English dog compared with dock, that is, vowels lengthen before a final voiced consonant, whereas in Balto-Slavic lengthening must have occurred before a non-final voiced consonant; cf. Lithuanian úoga or Old Church Slavonic vinjaga. Personally, we know of no language in which a change identical to the one Winter postulates for Balto-Slavic has taken place. And this raises a problem: is it possible that in Balto-Slavic a sound change occurred that has taken place in no other language?

3. There are cases whose interpretation seems doubtful.

Parmi les mots que cite J. H. Holst (2003: 164) pour étayer sa thèse, on trouve “lit. stógas; pbsl. *stágos < *stágos”. Il ne se rend pas compte qu’il existe une contradiction entre le lituanien et les langues slaves, où il y a des mots comme pol. stóg, tchèque stoh ou russe et slovène stog, qui proviennent de *stogъ. Un allongement, qui a eu lieu en lituanien, ne s’est pas produit dans les langues slaves.

Ailleurs, le même auteur (2003: 149) a écrit:

“Werner Winter has made the claim that a short vowel is lengthened before a voiced plosive in Balto-Slavic... whether it is correct, however, is highly controversial. This contribution shows that by modifying it, a correct claim can be made: the lengthening in question took place only under word stress. The fact that word stress plays a role in sound history is quite common. The best-known example is probably Verner’s Law.”

It must be emphasized that the vast majority of Germanists conceive of Verner’s law as Jespersen did, using two fictitious words: ásasa > ásaz(a) and asása > azás(a), that is, fricatives become voiced in medial and final position except when they follow the stressed vowel (Mańczak 1990: 92). Accordingly, all Germanists reconstruct Proto-Germanic words of the type

REMARQUES SUR LA RÈGLE DE WINTER

35

*wulfaz. Consequently, there is a difference between Winter’s rule, which, according to the modification proposed by Holst, concerns the position “unter dem Wortakzent”, and Verner’s rule, which applies to the position after a pretonic syllable and to that after a posttonic syllable.

4. For many years, we have maintained that, in all languages, the form of words depends on three principal factors: not only regular phonetic development and analogical development, but also what we call irregular phonetic development due to frequency. In all languages, frequently used morphemes, words, and word groups undergo irregular reductions. We have devoted three monographs to this subject (Mańczak 1969, 1977 and 1987), as well as numerous articles.

As for Winter’s rule, it has exceptions, which has led to divided opinions about it and prompted several attempts to modify it. We do not intend to discuss here whether this rule is valid, but would merely like to draw attention to the fact that certain exceptions to Winter’s rule could be explained by irregular phonetic development due to frequency.

V. sl. bogb. Alongside irregular phonetic development due to frequency, there is another kind of irregular change, consisting of phonetic accidents long known as assimilations, dissimilations, or metatheses, as well as hypercorrect forms or expressive forms. All this irregular change is characterized by the fact that it occurs in different languages in different words. For example, the Latin word caeruleus exhibits dissimilation, but it would be difficult to find in an Indo-European language a word meaning “caeruleus” that exhibits dissimilation. Unlike assimilations, dissimilations, etc., irregular phonetic development due to frequency occurs in different languages in a more or less parallel way, which is explained by the fact that the most frequent words are more or less the same everywhere. The name of God appears in a reduced form in various frequently used expressions: Polish, Russian, Slovak bodaj, dialectal Polish bdaj, Russian spasibo “thank you” < sъpasi Bogъ, Ukrainian spasybi, Czech bohdá (< Bůh dá), bodejt, bopomozi, dialectal Czech pánbiček (from the vocative pánbičku < pane na nebičku). The Slovak adjective bohovitý is derived from boh ho vie, except that an irregular reduction has occurred in the adjective. In addition, one may cite Slovak pomajbo “help, God”, pomodaj šťastia, Czech zdarbu < zďar Bůh, nazdar < nazdarďbůh, dialectal Polish dalbo < dalibóg, dialectal Polish pomagabóg, Lower Sorbian pomgaj Bog, Lower Sorbian lěč Bog da > zbogda > zboda, Ukrainian probi (= Polish przebóg) or bozna < *Bogъ znajett. The fricative pronunciation of g in Russian Bog is also explained by irregular phonetic development due to frequency (Mańczak 1997). Reduced forms also occur in Lithuanian: dieváž < dievaž(i) < Dievas žino(gi), sudie < sudieu < su Dievu, padėk < padėk Dieve, dialectal Lithuanian žegno(k) die < *žegno(k) Dieve. In Latin there are thousands of words beginning with d, but, as far as we know, Iuppiter is the only word that has lost its initial d. Obviously, this word, denoting the most important god in the Roman pantheon, was very frequently used. In our view, divus is a regular form, whereas

36 |

WITOLD MAŃCZAK

deus is explained by irregular phonetic development due to frequency (our paper on divus and deus will appear in the proceedings of a conference held in Łódź in 2004). In English, God be with you became good-bye. In American pronunciation, the word God has two pronunciations, not only with [ɑ], but also with [ɔ], whereas all the other words in -od (nod, rod, trod, etc.), which are less frequently used than God, have only [ɑ]. This is a reduction in the degree of vowel openness, which is characteristic of irregular phonetic development due to frequency (Mańczak 1995). In German, adieu was reduced to ade, whereas tschüs is an abbreviated form of Walloon adjuus, which is comparable to Spanish adiós < ad + Deus.

That the Slavic word bogъ “god” is a loanword is unlikely if we take into account the meaning of its derivatives, cf. Polish bogaty “rich”, ubogi “poor”, nieboga “unfortunate”, zboże “grain”, or Ukrainian bahato “a lot”.

V. sl. chodь. The verb meaning “to go” exhibits irregular reductions everywhere because it is frequently used. It is sufficient to note that the number-one problem in Romance etymology is the origin of verbs such as French aller, Italian andare, Spanish andar, Provençal ana, Romansh la, ma, na, etc. Since the sixteenth century, some sixty etymologies have been proposed to explain these forms, which in reality are nothing other than reduced forms of ambulare (Mańczak 1974 and 1975). Germanic verbs such as English go or German gehen have also undergone irregular phonetic development due to frequency (Mańczak 1987a). As for the Polish verb chodzić, it is interesting to note that its imperative is chodź, whereas the expected form would be *chódź (cf. the imperative wódź of wodzić “to lead”). The imperative chodź rather than *chódź is explained by the fact that in the Middle Ages, when Polish still had long vowels, the o of chodź underwent irregular shortening.

V. sl. voda. This frequently used word also exhibits irregular shortening. In Old French, Latin aqua regularly developed into aive, whereas the modern form eau is explained by irregular phonetic development due to frequency.

Lith. aš. The difference between Lith. aš and O. Sl. jazъ is also explained by development due to frequency.

V. sl. nevěsta. Au sujet de ce mot, Winter (1978: 444) a écrit ce qui suit:

“PIE wedh-... probably meant just ‘lead’. If the act of marriage involved not only the transfer of bridal presents, but also meant leading the bride away to the husband’s home (cf. Lat. in matrimonium ducere), a confusion of the two roots could easily enough happen: the more commonly used *wedh- was introduced in items where *(E)wed should have been kept, and as a result forms like OInd. vadhūs ‘bride’, Lith. vědinti ‘marry off’, etc., could come into existence.”

In our view, the most convincing etymology of nevěsta is *newo-wedh-tā “the newly brought home” and the reduction of *owe to ě is explained by irregular phonetic development due to frequency.

REMARQUES SUR LA RÈGLE DE WINTER

| 37

RÉFÉRENCES

HOLST, J. H. 2003. Eine Ausnahme des Winterschen Gesetzes. Historische Sprachforschung 116, 149–173

MAŃCZAK, W. 1969: Le développement phonétique des langues romanes et la fréquence. Kraków: Uniwersytet Jagielloński.

MAŃCZAK, W. 1974: Une étymologie romane controversée: aller, andar, etc. Revue roumaine de linguistique 19, 89–101.

MAŃCZAK, W. 1975: Étymologie de fr. aller, esp. andar, etc. et calcul des probabilités. Revue roumaine de linguistique 20, 735–739.

MAŃCZAK, W. 1977: Słowiańska fonetyka historyczna a frekwencja. Kraków Uniwersytet Jagielloński.

MAŃCZAK, W. 1987: Frequenzbedingter unregelmässiger Lautwandel in den germanischen Sprachen. Wrocław: Ossolineum.

MAŃCZAK, W. 1987a: Etymologie von gehen und stehen. Kwartalnik Neofilologiczny 34, 3–10.

MAŃCZAK, W. 1990: La restriction de la règle de Verner à la position médiane et le sort du s final en germanique. Historische Sprachforschung 103, 92–101.

MAŃCZAK, W. 1995: La prononciation américaine de God. Kwartalnik Neofilologiczny 42, 231–237.

MAŃCZAK, W. 1997: Frykatywne g w rosyjskim – rzekomy cerkiewizm. Słowianie Wschodni. Między językiem a kulturą, Kraków: Grell, 85–89.

WINTER, W. 1978 The distribution of short and long vowels in stems of the type Lith. ēsti: vēsti: mēsti and OCS jasti: vesti: mesti in Baltic and Slavic languages. Recent Developments in Historical Phonology, The Hague: Mouton, 431–446.

Witold Mańczak

Zakątek 13/59

PL–30–076 Kraków, Pologne